Hello Everybody,
not directly related to clojure .. but it is interesting to know that
scala people got a huge funding to further the state of the art in HPC...
http://www.scala-lang.org/node/8579
Sunil.
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient
with your first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
I have just finished
Language Virtualization for Heterogeneous Parallel Computing
http://infoscience.epfl.ch/record/148814/files/paper.pdf
The fundamental idea is defining a DSL (domain specific language)
to express some ideas. The DSL lives in a layer above the implementation
language but is embedded into the implementation language. That is,
the DSL does not have a separate compiler. DSLs define languages
that can support this as "virtualization languages" (VL). The VL needs
to provide four properties:
1) expressiveness -- everything should be expressible in the VL
2) performance -- the DSL should perform at VL speeds
3) safety -- the DSL should not reach into the underlying VL
4) effort -- the DSL should be easy to implement.
DSLs should be easy to combine because they both rest on the
same VL. Thus, a developer can provide capabilities of multiple
DSLs in the same image.
Clojure clearly fulfills all of the goals of a virtualization language
because it is a Lisp. Lisp systems are clearly capable of building
a DSL that is fully expressive of all of the power of Lisp. By using
the macro facility the DSL syntax can be matched to the problem
domain.
Macros allow the Clojure code to dynamically generate high performance
code that takes advantage of domain specific optimizations.
The macro language can be defined independent of the underlying
lisp so that there is no need to "reach below the VL" into Clojure.
This gives the DSL portability.
Macros are minimal effort. It is easy to develop macros and easy to
incrementally develop and enhance them to support any DSL features.
So I think that Clojure is an excellent virtualization language for DSLs.
While I do not know of any large Clojure applications yet, I know that
Axiom can be viewed as a mathematical domain specific language
in Common Lisp with facilities similar to Mathematica. ACL2 is a
logical domain specific language in Common Lisp with the ability
to prove lisp programs. Because they have a common VL they can
be combined into the same lisp image and ACL2 can be used to
prove Axiom mathematical statements. Lisp systems like Clojure
make this combination simple to implement.
I believe that, given enough time for Clojure to mature, we will see
large domain specific languages developed. These will be easy to
combine in a single Clojure image.
This paper provides a potential way to write parallel code for specific
domains. It attracted multi-million euro funding and could represent a
specific way forward for parallel Clojure programs.
Tim Daly
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
